Despite numerous attempts from providers, the cloud failed until now to secure a spot in the business intelligence market. However, there now is an increasingly pressing demand for it, particularly in the departmental applications area. The arrival of two new cloud specialists in the Gartner 2013 BI Magic Quadrant pinpoints the rise of those offerings.
Why did business intelligence remain on the sidelines of the cloud evolution? Was it a matter of time? We could reasonably think as much in light of recent developments, both in terms of supply and demand. The market seems to be opening up to three sub-segments of business intelligence: performance management applications, which was the focus of the first article in this series; departmental BI applications, which will be discussed in this article; and, finally, the enterprise data warehouse and big data, which will be the subject of the concluding article in this series.
If there was ever proof of cloud BI’s worth, it is web analytics for clickstream analysis. According to Gartner, it would seem that 98% of such applications are deployed in the cloud. Web analytics
is an ideal match for deployment in the cloud since data is sourced from outside the company, its format is standardized and the company usually wishes to cross-reference data gathered through its own website with external data from the Web.
This success has led some of the BI market players to position themselves in analytics at the intersection of departmental BI and performance management applications. At the core of their solutions is a library of key performance indicators. Those indicators are categorized by activities and are often associated with connectors that source the content needed to calculate the value of those indicators from popular-off-the shelf enterprise applications packages, such as SAP, Oracle Applications or Salesforce.com. Some examples of such players are Mirror42 (offering a library of more than 6,000 indicators on the KPIlibrary.com site) and GoodData. Also, IBM recently announced Analytic Answers, a predictive analytics
offering for small and medium-sized companies, with sector ranges such as insurance, retail or education.
The web analytics example has shown the appeal of the cloud model when data is sourced within the cloud itself. For the same reasons, cloud-based analytics solutions are particularly well suited to those whose enterprise applications are in the cloud. Providers of these solutions in SaaS mode are thus potentially ideally placed to combine them with analytical applications. But, as is the case for the SaaS world leader, Salesforce.com, their offerings are sometimes weak in the analytics area, which leaves a white space for third-party solutions.
More Agile "Departmental" BI Applications
The underlying infrastructure needed to deliver on the promises of business intelligence is still a primary concern in many organizations. Indeed, legacy architectures struggle to cope with the exploding volume of data and the ever-increasing user demand for empowerment. New technologies, such as in memory, do bring relevant solutions for this type of problem, but they require a constant readjustment of the infrastructure to suit uses. As opposed to transactional environments in which demand forecasts are controlled, BI demand is more volatile and challenging to predict.
Moreover, business intelligence requires considerable agility. There is currently a high demand for the setup of "data labs" to satisfy the individual needs of small groups of users over a sometimes short period of time, or to extend the reach of existing systems to a wider population. Similarly, it is often necessary to recover data in more detail or more frequently, and this may disrupt existing infrastructures. Standardization has caused business intelligence to lose some agility, and this is mostly due to its underlying technological infrastructure.
The cloud model, and its renowned flexibility, is particularly suited to this kind of demand. Some players, such as BIME or Birst, specialize in cloud solutions.
It should be noted that this sub-segment, as with performance management, is not necessarily limited to players positioned exclusively in the SaaS market. Traditional enterprise software vendors have recently clarified their cloud strategy, including for BI. MicroStrategy, for instance, provides a cloud version of its tools enriched by third-party databases and integration tools. SAP
just released a cloud version of its latest BI self-service offering called Lumira and partnered with Amazon to make HANA accessible through the cloud with a pay-per-use model – an offering that unfortunately is limited (with a database capacity of only a few dozens of gigabytes), but nonetheless appealing if only to jump-start new projects. Oracle, during its annual convention, OpenWorld, excitedly announced its new database version, now better suited to the cloud and its flexible nature. Similarly, at its 2013 Sapphire conference, SAP announced Hana Enterprise Cloud, a complete platform to run both analytical and transactional applications, including “legacy” SAP applications such as BW or ECC, off premises in a cloud environment. And Microsoft, even though business intelligence is not a pillar of its Azure platform as a service, is also progressively entering in the game, as shown by the availability of Windows Azure SQL Reporting in S2 2012.
Following the same line of reasoning, some hosting or cloud specialists are providing standardized, dedicated solutions for some of the market's BI platforms. At Business & Decision, we used our Eolas subsidiary’s infrastructure as a foundation to design our Qloud Services offering for QlikView as well as dedicated offerings for SAP and Oracle planning and financial consolidation platforms.
This model’s limits? If source data is not already located within the cloud, companies may be reluctant to transport (and eventually store, depending on the cloud BI architecture) it externally. This is one limitation that must be closely examined and dealt with on a case-by-case basis. Most service providers know how to integrate a virtual private network to take care of that problem. Another limitation is the volume of data to be transferred, which explains why this sub-segment should be positioned for departmental BI applications or for enterprise BI in companies that do not have to deal with massive volumes of data or do not have to transfer data that is too complex. Finally, the last limitation relates to the vendor’s license model: if the cloud offering is provided through a third party and not by the vendor itself and if this third party cannot offer the software by subscription as part of its services, the company must buy a perpetual license in order to use the service. Expenses must then be treated as an investment (CAPEX) and not as solely operation-related (OPEX), and this can become a constraint in terms of project financing. Additionally, the third-party service may not include automatic application of release notes or upgrades to the latest version.
The concluding article in this series will look at enterprise business intelligence and big data in the cloud.
Recent articles by Jean-Michel Franco